Domain knowledge graph construction method and system based on graph neural network

Abstract

The invention provides a domain knowledge graph construction method and system based on a graph neural network, and relates to the field of knowledge maps, and the method comprises the steps: obtaining basic data of a target domain, carrying out the data analysis of the basic data, and extracting knowledge nodes and a node path composed of the knowledge nodes from the basic data; dividing a plurality of target domain node groups according to the context association features of the knowledge nodes and the matching degree of the knowledge nodes and the target domain; the method comprises the following steps: mining potential association paths among knowledge nodes in a target domain node group, constructing a knowledge graph, determining a core node in the knowledge graph, obtaining preliminary push content from a preset database according to the core node, and determining a push priority for the preliminary push content according to the preference of a user to obtain priority push content. According to the method, the association degree of the knowledge graph and the target domain is accurately evaluated, meanwhile, the core concept and the key relation are accurately recognized, and intelligent and personalized knowledge pushing is achieved.

Description

Technical Field

[0001] This invention relates to the field of knowledge graphs, and more particularly to a method and system for constructing a domain knowledge graph based on graph neural networks. Background Technology

[0002] In the information age, building domain-specific knowledge graphs and achieving precise knowledge delivery is particularly crucial. This is not only a vital cornerstone for promoting intelligent information processing but also a core support for improving users' efficiency in obtaining effective information. By structuring and connecting fragmented knowledge, decision support and in-depth insights can be provided to various industries.

[0003] However, current knowledge graphs have significant shortcomings in practical applications. Their application is still limited to simple knowledge storage and retrieval, lacking in-depth exploration of the correlation between knowledge and target domains, and making it difficult to accurately judge the importance of knowledge. As a result, the content pushed often does not meet user needs and is inefficient.

[0004] Therefore, how to strengthen the connection between knowledge domains and target domains based on the construction of knowledge graphs, accurately identify core concepts and key relationships, and thereby optimize the response speed of frequently accessed content is a key problem that the industry urgently needs to solve. Summary of the Invention

[0005] In view of this, the present invention proposes a method and system for constructing a domain knowledge graph based on graph neural networks, thereby solving the problems that the constructed knowledge graph lacks in-depth mining of the correlation between knowledge and the target domain, and is also difficult to accurately judge the importance of knowledge.

[0006] The technical solution of this invention is implemented as follows: According to a first aspect, embodiments of the present invention provide a method for constructing a domain knowledge graph based on a graph neural network, the method comprising: Acquire basic data in the target domain, perform data analysis on the basic data, and extract knowledge nodes and node paths composed of knowledge nodes from the basic data; the knowledge nodes form node paths through logical relationships. Based on the contextual association features of knowledge nodes and their matching degree with the target domain, several target domain node groups are divided; each target domain node group includes a central node and knowledge nodes that have a logical relationship with the central node; The system mines potential connections between knowledge nodes in a target domain node group, constructs a knowledge graph based on knowledge nodes, logical relationships, and potential connections, identifies core nodes in the knowledge graph, obtains initial push content from a pre-set database based on the core nodes, and determines the push priority of the initial push content based on user preferences to obtain priority push content.

[0007] In conjunction with the first aspect, in the first embodiment of the first aspect, the step of dividing a number of target domain node groups according to the contextual association features of knowledge nodes and their matching degree with the target domain specifically includes: Determine the matching degree between knowledge nodes and the target domain, and retain knowledge nodes whose matching degree exceeds the first preset matching degree; Determine the contextual association features of knowledge nodes, identify the central node based on the contextual association features, and divide the candidate node groups according to the logical relationships of the central nodes; The contextual features of each knowledge node in each candidate node group are fused to obtain node fusion features. The matching degree between the node fusion features and the target domain is determined, and the candidate node group with a matching degree exceeding the second preset matching degree is taken as the target domain node group.

[0008] In conjunction with the first implementation method of the first aspect, in the second implementation method of the first aspect, the step of mining potential connection paths between various knowledge nodes in the target domain node group, constructing a knowledge graph based on knowledge nodes, logical relationships, and potential connection paths, determining core nodes in the knowledge graph, obtaining preliminary push content from a preset database based on the core nodes, and determining the push priority of the preliminary push content based on user preferences to obtain priority push content specifically includes: Obtain the semantic similarity and context dependency matrix among the knowledge nodes in the target domain node group; Based on semantic similarity and context dependency matrix, and using graph neural network model, the graph neural network model outputs the connection strength between each knowledge node in the target domain node group, and the potential association path between knowledge nodes is mined based on the connection strength. Based on the logical relationships and potential association paths between knowledge nodes in all target domain node groups, an initial framework for the knowledge graph is constructed, and the centrality of each knowledge node in the initial framework is determined based on the connection strength and the number of connections. Knowledge nodes with centrality exceeding the preset centrality are designated as core nodes, and initial push content is obtained from the preset database based on the core nodes; The system acquires user interaction data within a first preset time period, determines the frequency of user interaction with each type of initial push content based on the interaction data, and determines user profile data representing user behavior patterns based on the interaction frequency. The interaction data includes the user's historical data and real-time behavior data. The push priority of the initial push content is determined based on the profile data. The initial push content is pushed to the user according to the push priority, and the user's interaction data is obtained in the second preset time period. By comparing the interaction data of the first and second time periods, the user's preferred content is obtained, and the push priority is adjusted according to the preferred content. The priority is then determined based on the adjusted priority.

[0009] In conjunction with the second embodiment of the first aspect, in the third embodiment of the first aspect, before the step of designating knowledge nodes with centrality exceeding a preset centrality as core nodes and obtaining preliminary push content from a preset database based on the core nodes, the method further includes: Adjust potential related paths according to the target domain; the adjustment methods include adding potential related paths and deleting potential related paths.

[0010] In conjunction with the first aspect, in the fourth embodiment of the first aspect, the method further includes: Priority-driven content is stored using distributed storage, and cache space is allocated to priority-driven content based on its priority; higher-priority content receives more cache space.

[0011] In conjunction with the fourth embodiment of the first aspect, in the fifth embodiment of the first aspect, the method further includes: Obtain access request logs for each priority level of pushed content, extract access distribution characteristics from the access request logs, and adjust the cache space allocated to priority-pushed content based on the access distribution characteristics.

[0012] In conjunction with the fifth embodiment of the first aspect, in the sixth embodiment of the first aspect, the method further includes: Based on the adjusted storage space for priority-pushed content, the system retrieves the runtime data of the knowledge nodes corresponding to the priority-pushed content and then distributes access requests for the priority-pushed content according to this runtime data. This runtime data includes response time and resource utilization.

[0013] In conjunction with the sixth embodiment of the first aspect, in the seventh embodiment of the first aspect, the method further includes: Based on the access request logs of the target domain and the content pushed for each priority, the logical relationships between knowledge nodes are reconstructed, and the initial architecture of the knowledge graph is adjusted.

[0014] According to a second aspect, embodiments of the present invention provide a domain knowledge graph construction system based on graph neural networks, the system comprising: The data acquisition module is used to acquire basic data in the target domain, perform data analysis on the basic data, and extract knowledge nodes and node paths composed of knowledge nodes from the basic data; the knowledge nodes are connected by logical relationships to form node paths. The data matching module is used to divide several target domain node groups according to the contextual association features of knowledge nodes and their matching degree with the target domain; each target domain node group includes a central node and knowledge nodes that have a logical relationship with the central node; The knowledge graph construction module is used to mine the potential connection paths between various knowledge nodes in the target domain node group. It constructs a knowledge graph based on knowledge nodes, logical relationships, and potential connection paths, identifies the core nodes in the knowledge graph, obtains preliminary push content from a preset database based on the core nodes, and determines the push priority of the preliminary push content based on user preferences to obtain priority push content.

[0015] The domain knowledge graph construction method and system based on graph neural networks of the present invention have the following advantages over the prior art: After acquiring knowledge nodes, several target domain node groups are formed based on their contextual features and matching degree with the target domain. These target domain node groups are highly relevant to the target domain, ensuring a high degree of alignment between the knowledge content and the needs of the target domain. Potential connection paths between knowledge nodes within each target domain node group are then mined to construct a knowledge graph. Core nodes in the knowledge graph are determined based on the centrality of the knowledge nodes. Initial content is pushed from a pre-set database based on these core nodes. This approach accurately identifies core concepts and key relationships, enhances the system's adaptability in specific scenarios, and improves the application relevance of knowledge nodes. Acquiring user preferences allows for timely capture of dynamic changes in user interests, enabling adjustments to push priorities based on these preferences. Furthermore, an efficient storage architecture optimizes the response speed for frequently accessed content, achieving intelligent and personalized knowledge push. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is one of the flowcharts illustrating the domain knowledge graph construction method based on graph neural networks of the present invention; Figure 2 This is the second flowchart illustrating the domain knowledge graph construction method based on graph neural networks of the present invention. Figure 3 This is the third flowchart illustrating the domain knowledge graph construction method based on graph neural networks of the present invention; Figure 4 This is the fourth flowchart illustrating the domain knowledge graph construction method based on graph neural networks of the present invention. Figure 5 This is a schematic diagram of the domain knowledge graph construction system based on graph neural networks according to the present invention. Detailed Implementation

[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] In the information age, building domain-specific knowledge graphs and achieving precise knowledge delivery is particularly crucial. This is not only a vital cornerstone for promoting intelligent information processing but also a core support for improving users' efficiency in obtaining effective information. By structuring and connecting fragmented knowledge, decision support and in-depth insights can be provided to various industries.

[0020] However, current knowledge graphs have significant shortcomings in practical applications. Their application is still limited to simple knowledge storage and retrieval, lacking in-depth exploration of the correlation between knowledge and target domains, and making it difficult to accurately judge the importance of knowledge. As a result, the content pushed often does not meet user needs and is inefficient.

[0021] The main reasons for the above problems are as follows: assessing the relevance of knowledge to the target domain has become the primary challenge. Due to the lack of an effective measurement mechanism, the system has difficulty in determining which knowledge truly fits the needs of a specific scenario. In the complex knowledge system, the identification of core concepts and key relationships has become extremely difficult, making it impossible to accurately extract the most valuable knowledge points. Because important knowledge has not been effectively screened and integrated, the system often faces the dilemma of uneven resource allocation and slow response speed when processing frequently accessed content, which affects the user experience.

[0022] In summary, how to strengthen the connection between knowledge domains and target domains based on the construction of knowledge graphs, accurately identify core concepts and key relationships, and thereby optimize the response speed of frequently accessed content is a key problem that the industry urgently needs to solve.

[0023] The domain knowledge graph construction method based on graph neural networks provided in this manual aims to accurately assess the correlation between the knowledge graph and the target domain, accurately identify core concepts and key relationships, and optimize the response speed of frequently accessed content through an efficient storage architecture, thereby realizing intelligent and personalized knowledge push, improving the push effect and system operating efficiency.

[0024] Please see Figure 1 , Figure 1 This is a flowchart illustrating the domain knowledge graph construction method based on graph neural networks of the present invention. The method may include the following steps: S101. Obtain basic data in the target domain, perform data analysis on the basic data, and extract knowledge nodes and node paths composed of knowledge nodes from the basic data.

[0025] In this embodiment, the basic data mentioned above is obtained through a pre-established domain knowledge base. The domain knowledge base can be supplemented with the basic data of the target domain stored therein by continuously collecting and summarizing relevant literature, records and interview data.

[0026] More specifically, step S101 includes: S1011. Obtain basic data in the target domain and convert the basic data into text data.

[0027] For example, if the basic data includes data in the form of tables, text, images, and audio, it is first converted into text data in a unified way. This helps with subsequent semantic analysis, reduces ambiguity, and improves the accuracy of data processing.

[0028] S1012. Preprocess the text data and then perform word segmentation on the preprocessed text data to obtain text fragments.

[0029] Text data often contains invalid characters or inconsistent formatting. In this embodiment, preliminary preprocessing, or data cleaning, is performed on the basic data before data analysis. For example, general noise reduction tools are used to remove redundant spaces, special symbols, or irrelevant annotations from the text data to ensure basic readability. The converted text data is then segmented and split into several text fragments, i.e., a set of text fragments.

[0030] S1013. Perform semantic analysis on the text fragments to determine the contextual relevance between them, and classify text fragments with a relevance exceeding a preset threshold into the same knowledge node. The knowledge nodes are connected by logical relationships to form node paths.

[0031] Semantic analysis can be used to determine the semantic understanding information of all text fragments. Based on this information, the contextual relevance between different text fragments can be obtained. If the contextual relevance between two text fragments exceeds a preset threshold, these two text fragments can be categorized into a structured knowledge unit group. If any subsequent text fragment has a contextual relevance with a text fragment in this knowledge unit group that exceeds the preset threshold, that text fragment will also be categorized into that knowledge unit group. In this way, all text fragments can be divided into several knowledge unit groups, each of which is a corresponding knowledge node. The categorized knowledge unit groups provide a clear logical foundation for the construction of knowledge nodes. This process effectively integrates scattered information into meaningful knowledge blocks, facilitating subsequent applications.

[0032] S1014. Determine the logical relationships between different knowledge nodes, and use the logical relationships as node connection information to connect knowledge nodes. Based on the node connection information, establish a node path composed of knowledge nodes.

[0033] By mapping knowledge nodes together, they can be integrated, connecting logically related nodes into node paths. Different node paths can be connected by at least one knowledge node, and different node paths can also share the same knowledge node. For example, knowledge nodes B, C, and D are connected to form node path A, while knowledge nodes A, B, E, and F are connected to form node path B. Node paths A and B share the same knowledge node B. Knowledge nodes F, G, H, I, J, and K are connected to form node path C. Node paths B are connected to each other through knowledge node F, which is also a shared knowledge node between node paths B and C.

[0034] The logical relationships between knowledge nodes are key information connecting two different knowledge nodes. For example, knowledge node B and knowledge node C are connected through node connection information "extraction patterns." This node connection information also serves as prompts / retrieval information for the node paths formed by these knowledge nodes, creating a complete chain extending from one piece of information to others. This associative mapping not only improves the traceability of knowledge but also provides data support for subsequent intelligent question answering or decision support systems. When users search for relevant information in a target domain, they can quickly find the content they need through knowledge nodes, reducing search time. Simultaneously, structured knowledge units and node sets provide a reliable data foundation for AI-assisted processing, significantly improving the system's intelligence level.

[0035] It should be noted that multiple knowledge nodes can establish a logical relationship with knowledge node A, such as knowledge node B, knowledge node C, and knowledge node D. However, not all knowledge nodes have a logical relationship with each other; for example, knowledge node D and knowledge node I may not have a logical relationship.

[0036] S102. Based on the contextual association features of knowledge nodes and their matching degree with the target domain, several target domain node groups are divided, wherein each target domain node group includes a central node and knowledge nodes that have a logical relationship with the central node.

[0037] By calculating the matching degree between knowledge nodes and the target domain, it is possible to ensure that the selected knowledge nodes are highly relevant to the core needs. This forms a complete logical chain from the association calculation of knowledge nodes to the final division and grouping, ensuring that the knowledge content is highly consistent with the needs of the target domain and laying a solid foundation for subsequent applications.

[0038] S103. Mine potential connection paths between knowledge nodes in the target domain node group, construct a knowledge graph based on knowledge nodes, logical relationships, and potential connection paths, determine the core nodes in the knowledge graph, obtain preliminary push content from the preset database based on the core nodes, and determine the push priority of the preliminary push content based on user preferences to obtain priority push content.

[0039] By analyzing the potential connections between knowledge nodes, we can focus on core relationships and avoid interference from irrelevant information. Core nodes are those with high centrality in the knowledge graph. Centrality allows us to quickly filter out node information that is crucial for knowledge delivery. Acquiring user preferences allows us to capture dynamic changes in user interests in a timely manner, enabling us to adjust the delivery priority based on these preferences, thus achieving prioritized content delivery.

[0040] The domain knowledge graph construction method based on graph neural networks of this invention, after acquiring knowledge nodes, divides them into several target domain node groups based on the contextual association features of the knowledge nodes and their matching degree with the target domain. These target domain node groups are highly correlated with the target domain, ensuring a high degree of fit between the knowledge content and the needs of the target domain. Then, the potential association paths between the knowledge nodes in each target domain node group are mined to construct the knowledge graph. The core nodes in the knowledge graph are determined based on the centrality of the knowledge nodes. Initial push content is obtained from a preset database based on the core nodes. This method can accurately identify core concepts and key relationships, enhance the system's adaptability in specific scenarios, improve the application targeting of knowledge nodes, and capture the dynamic changes in user interests in a timely manner. This allows for the adjustment of push priority based on user preferences to obtain prioritized content. Furthermore, an efficient storage architecture optimizes the response speed of frequently accessed content, achieving intelligent and personalized knowledge push.

[0041] Please see Figure 2 The method may also include the following steps: S201. Obtain basic data for the target domain, perform data analysis on the basic data, and extract knowledge nodes and node paths composed of knowledge nodes from the basic data. Refer to step S101 for details.

[0042] S2021. Determine the matching degree between knowledge nodes and the target domain, and retain knowledge nodes whose matching degree exceeds the first preset matching degree.

[0043] In this embodiment, graph structure tools can be used to evaluate the connection relationships (i.e., whether they are connected and their logical relationships) between knowledge nodes, analyze the connection paths and frequencies with the target domain requirements, and then obtain the matching degree between knowledge nodes and the target domain. Knowledge nodes with a matching degree exceeding the preset matching degree are retained, while knowledge nodes with a matching degree lower than the preset matching degree are deleted. This ensures that the selected knowledge nodes are highly relevant to the core requirements.

[0044] S2022. Determine the contextual association features of knowledge nodes, identify the central node based on the contextual association features, and divide the candidate node groups according to the logical relationship of the central node.

[0045] In this embodiment, contextual association features refer to the surrounding connection environment of a knowledge node in the node path, that is, the knowledge nodes that have a logical relationship with a certain knowledge node and the number of these knowledge nodes. Since the knowledge nodes have been initially screened in step S2021, the path relationship is analyzed based on the contextual association features of the retained knowledge nodes in step S2022. For example, if knowledge node A has a shorter path with knowledge node B, knowledge node L, etc., the weight value of the contextual association feature of knowledge node A is higher, and the importance of knowledge node A in the target domain requirement is confirmed. Knowledge node A is determined as the central node (the central node is also a knowledge node). This analysis helps to identify the actual role of the node in a specific scenario.

[0046] Then, based on the identified central node, knowledge nodes that have a logical relationship with the central node, along with the central node itself, are grouped into candidate node groups. For example, a group of knowledge nodes surrounding knowledge node A forms candidate node group A, and a group of knowledge nodes surrounding knowledge node B forms candidate node group B. This grouping method can focus on the core content within the domain.

[0047] It should be noted that the same knowledge node can be divided into multiple candidate node groups.

[0048] S2023. Perform feature fusion on the contextual association features of each knowledge node in each candidate node group to obtain node fusion features, determine the matching degree between the node fusion features and the target domain, and take the candidate node group with the matching degree exceeding the second preset matching degree as the target domain node group.

[0049] In this embodiment, the value of the second preset matching degree exceeds the value of the first preset matching degree, thus enabling deeper filtering. In this embodiment, all target domain node groups can also be prioritized based on node fusion features. The higher the matching degree between the node fusion features and the target domain, the higher the priority of the target domain node group. This sorting mechanism ensures that the most relevant knowledge content is presented first, improving the targeting of knowledge management.

[0050] S303. Mine potential connection paths between knowledge nodes in the target domain node group, construct a knowledge graph based on knowledge nodes, logical relationships, and potential connection paths, determine the core nodes in the knowledge graph, obtain preliminary push content from a preset database based on the core nodes, and determine the push priority of the preliminary push content according to user preferences to obtain priority push content. Refer to step S103 for details.

[0051] Please see Figure 3 The method may also include the following steps: S301. Obtain basic data for the target domain, perform data analysis on the basic data, and extract knowledge nodes and node paths composed of knowledge nodes from the basic data. Refer to step S101 for details.

[0052] S302. Based on the contextual association features of knowledge nodes and their matching degree with the target domain, divide them into several target domain node groups. For details, refer to step S102.

[0053] S3031. Obtain the semantic similarity and context dependency matrix between each knowledge node in the target domain node group.

[0054] For example, in a knowledge management platform, a target domain node group contains multiple knowledge nodes related to "intelligent recommendation system," such as "user preference" nodes, "content tag" nodes, and "recommendation algorithm" nodes. Semantic analysis reveals that the semantic similarity between the "user preference" node and the "content tag" node is 0.85, while the semantic similarity between the "user preference" node and the "recommendation algorithm" node is only 0.45. Furthermore, contextual dependency data indicates that the "user preference" node frequently appears within the context of the "content tag" node, forming a strong contextual dependency. This quantitative processing provides foundational data for subsequent association mining.

[0055] S3032. Based on semantic similarity and context dependency matrix and using graph neural network model, output the connection strength between each knowledge node in the target domain node group by graph neural network model, and mine the potential association path between knowledge nodes based on the connection strength.

[0056] By inputting semantic similarity and context dependency matrices into a pre-trained graph neural network model, the connection strength between a knowledge node and other knowledge nodes can be obtained from the graph neural network model.

[0057] It should be noted that if a knowledge node already has a logical relationship with another knowledge node, the connection strength reflects the correlation between the knowledge nodes; if a knowledge node does not have a logical relationship with another knowledge node, the connection strength is the possible node correlation between the two nodes.

[0058] Next, connection strengths exceeding a preset value are identified as potential association paths. For example, using a graph neural network model, the connection strength from the "User Preferences" node to the "Content Tag" node is 0.9, higher than the preset value of 0.6, therefore this path is identified as a potential association path. The connection strength from the "User Preferences" node to the "Recommendation Algorithm" node is only 0.5, which does not meet the preset value and is temporarily excluded. Mining potential association paths between knowledge nodes by strength helps to focus on core associations and avoid interference from irrelevant information.

[0059] S3033. Adjust potential related paths according to the target domain.

[0060] In this embodiment, the adjustment methods include adjusting the connected knowledge nodes, such as adding or deleting potential related paths. Based on the target domain, the contextual logic relationship between two knowledge nodes with potential related paths can be verified. For example, if the contextual logic of the potential related path between the "user preference" node and the "content tag" node shows that they often work together to improve user experience in recommendation scenarios, then the verification passes. Another potential related path might be eliminated due to unclear contextual logic. Simultaneously, if the target domain reveals a high degree of correlation between two knowledge nodes, a potential related path can be established between these two nodes.

[0061] By aligning the path with the target domain, the practical application value of the path is ensured. This multi-dimensional analysis ensures the comprehensiveness of the results, covering not only the core elements of the recommendation system but also reflecting the logical dependencies between nodes through path associations. This provides a solid foundation for subsequent knowledge applications, enhances the relevance, rationality, and adaptability of knowledge nodes to actual needs, and strengthens the system's adaptability in specific scenarios.

[0062] S3034. Based on the logical relationships and potential association paths between knowledge nodes in all target domain node groups, construct the initial framework of the knowledge graph, and determine the centrality of each knowledge node in the initial framework based on the connection strength and the number of connections.

[0063] After obtaining the logical relationships and potential association paths between each knowledge node in the target domain node group, an initial framework of the knowledge graph can be constructed based on this information. After processing in step S3032, the logical relationships (also potential association paths) and potential association paths have corresponding connection strengths. By assigning certain computational weights to the connection strength of each path and the number of other knowledge nodes connected, the centrality of each knowledge node in the initial framework can be determined.

[0064] S3035. Select knowledge nodes with centrality exceeding the preset centrality as core nodes, and obtain initial push content from the preset database based on the core nodes.

[0065] When the calculated centrality of a knowledge node, such as the "Machine Learning" node, exceeds a preset centrality, that knowledge node is marked as a core node. This allows for quick filtering of node information that is crucial for knowledge delivery. Next, preliminary delivery content related to the "Machine Learning" node is retrieved. Assuming the preset database contains the latest academic paper abstracts, tutorial resources, etc., knowledge content directly related to the "Machine Learning" node is extracted to form structured preliminary delivery content. For example, the content might be divided into two categories: "Basic Theory" and "Application Cases." Retrieving this preliminary delivery content facilitates more targeted distribution later.

[0066] In this embodiment, all core nodes can also be prioritized based on centrality. According to the specific value of centrality, the higher the centrality of the core node, the higher its priority, and the higher the priority of its corresponding initial push content. In this way, the initial push content not only covers the core knowledge points, but also the sorting of the push content can improve the efficiency and convenience of users to acquire knowledge, while ensuring the relevance and practicality of the content.

[0067] The initial push content can then be processed according to its priority order.

[0068] S3036. Obtain user interaction data within a first preset time period, determine the user's interaction frequency with each type of initial push content based on the interaction data, and determine profile data representing the user's behavior pattern based on the interaction frequency. The interaction data includes the user's historical data and real-time behavior data.

[0069] In this embodiment, the first preset time period is a specified period of time, such as one month. By obtaining the user's historical data and real-time behavior data within the preset time period, the interaction frequency between the user and each type of initial push content can be clearly obtained.

[0070] For example, a user's frequent search history for "deep learning" related topics in the past month, as well as their recent clicks on articles related to "neural networks," can be used to construct a profile of the user's behavioral patterns based on the frequency of interaction. This profile might show that the user's interest in the field of "deep learning" accounts for 60%, while their interest in other branches such as "data mining" is only 20%.

[0071] S3037. Determine the push priority of the initial push content based on the profile data, push the initial push content to the user according to the push priority, and obtain the user's interaction data in the second time period.

[0072] Since the profile data is obtained based on the user's interaction frequency, the push priority of all initial push content can be determined from this. Then, the initial push content is pushed to the user according to the push priority, and the user's interaction data of the content in the second preset time period is obtained. The interaction data of the content in the second preset time period is the user's preference feedback data in the second preset time period.

[0073] S3038. By comparing the interaction data of the first time period and the second time period, the user's preferred content is obtained, and the push priority is adjusted according to the preferred content. The priority is then adjusted based on the initial adjustment content to obtain the priority push content.

[0074] In this embodiment, the second preset time period is also a specified period of time, such as one week. The interaction data of the content in the second preset time period can be as follows: after the initial content is pushed, the user clicks on the first two "application cases" content and stays on the platform for more than 5 minutes, and also saves a related article. By comparing the historical records, it is found that the user's attention to "application cases" has increased by 30% recently compared with before, indicating that there is a change in behavior pattern, which reflects the user's preferred content.

[0075] By capturing user preferences, we can promptly identify dynamic changes in user interests. This allows us to adjust push notification priorities based on these preferences, resulting in prioritized content. Understandably, prioritized content also has a corresponding priority level; higher-priority content becomes the primary focus of push notifications. This prioritization method ensures targeted content delivery, guaranteeing that the content better matches the user's current needs, while also improving the efficiency of information retrieval through dynamic adjustments.

[0076] Please see Figure 4 The method may also include the following steps: S401. Obtain basic data for the target domain, perform data analysis on the basic data, and extract knowledge nodes and node paths composed of knowledge nodes from the basic data. Refer to step S101 for details.

[0077] S402. Based on the contextual association features of knowledge nodes and their matching degree with the target domain, divide them into several target domain node groups. For details, refer to step S102.

[0078] S403. Mine potential connection paths between knowledge nodes in the target domain node group, construct a knowledge graph based on knowledge nodes, logical relationships, and potential connection paths, determine the core nodes in the knowledge graph, obtain preliminary push content from a preset database based on the core nodes, and determine the push priority of the preliminary push content according to user preferences to obtain priority push content. Refer to step S103 for details.

[0079] S404. Priority push content is stored through distributed storage, and cache space is allocated to priority push content according to its priority. Higher priority push content receives more cache space.

[0080] In this embodiment, a cache-based data storage framework is provided for the final priority push content. This data storage framework allocates more cache space for high-priority push content and optimizes the data reading path through a distributed storage mechanism.

[0081] Specifically, distributed storage tools are used to partition the cache space of the distributed storage system. Content is pushed based on priority; higher-priority content is allocated to higher-priority areas. This adjusted allocation ensures that frequently accessed content is quickly retrieved, reducing user wait time. Based on the adjusted allocation, data read path information is obtained, and path optimization tools are used to dynamically adjust this information. For frequently accessed content, the path length is shortened to obtain an optimized path scheme. For example, if the original path required multiple node jumps and was time-consuming, the path optimization tool shortens the path length from 5 nodes to 2 nodes. This adjustment allows users to load resources faster when accessing frequently accessed priority content, improving access efficiency. For the optimized path scheme, the storage location of priority push content is finally adjusted through storage management tools. Combining the characteristics of distributed storage, it is determined whether it meets the access frequency requirements, and the final storage configuration is obtained. It is assumed that a certain type of relatively high-frequency priority push content is assigned to a storage node closer to the user, while a certain type of relatively low-frequency priority push content is stored in a more distant node. The final storage configuration is then confirmed. This configuration method can effectively balance the use of storage resources while ensuring fast response of high-frequency content.

[0082] Preferably, the data storage framework can also be optimized according to the distribution of the user group. Assuming that most users are concentrated in a certain area, the data storage framework will prioritize storing relevant priority push content on nodes near that area to ensure access speed.

[0083] S405. Obtain the access request logs for each priority type of pushed content, extract the access distribution characteristics from the access request logs, and adjust the cache space allocated to the priority pushed content based on the access distribution characteristics.

[0084] In this embodiment, detailed records of access request logs for each priority type of pushed content are obtained from the knowledge graph's operation log. Then, access distribution features are extracted from the access request logs. These access distribution features represent the actual access frequency distribution data of the priority-push content that has been allocated priority and cache space. In this way, the cache space is dynamically readjusted for each priority type of pushed content based on the actual access frequency distribution data.

[0085] When dynamically adjusting cache space, optimization can be based on the current cache usage. For example, if the total cache space is 100GB, with 40GB occupied by frequently accessed priority content, but based on access distribution characteristics, the cache space for frequently accessed priority content should be increased to 60GB. The storage management tool will then reallocate the cache space ratio accordingly, compressing the cache space for infrequently accessed priority content while expanding the storage area for frequently accessed priority content. This dynamic cache space adjustment method ensures the rational utilization of physical resources.

[0086] Preferably, when implementing the dynamic adjustment of cache space, i.e. resource allocation scheme, into the real-time strategy, the changes in access request logs can be recorded by real-time monitoring tools. The logging tool will continuously track this change, and if it finds that the access request logs suddenly drop, it will report back to the system and trigger a new round of dynamic adjustment of cache space.

[0087] S406. Based on the adjusted storage space according to the priority of the pushed content, obtain the runtime data of the knowledge nodes corresponding to the priority pushed content, and perform traffic distribution processing on the access requests for the priority pushed content according to the runtime data. The runtime data includes response time and resource utilization.

[0088] In this embodiment, the operation data of each knowledge node is monitored in real time. The operation data can reflect whether the knowledge node is a high-load node. If a knowledge node is determined to be a high-load node, the access requests to the knowledge node are distributed using a load balancing tool. Combined with the request distribution logic, some requests are transferred from the high-load node to nodes with lower resource utilization, avoiding overload of the high-load node and resulting in response delay. This ensures that the knowledge node can quickly adapt to load changes and guarantees the smooth operation of the system.

[0089] Understandably, after the traffic is split, the runtime data of the knowledge nodes corresponding to the priority content will be continuously monitored. If the runtime data of the knowledge node indicates that the knowledge node is in a relatively balanced state of response time and resource utilization, it means that the adjustment is successful and no further adjustment is needed for the time being.

[0090] In this embodiment, after the traffic splitting process, the user's access request logs are also obtained. Then, based on the access distribution characteristics in the access request logs, steps S405 to S406 are processed cyclically to optimize the knowledge graph structure until the priority push content dynamically adapts to changes in user needs, ensuring the long-term attractiveness of the recommended content.

[0091] S407. Based on the access request logs of the target domain and the content pushed for each priority, reconstruct the logical relationships between knowledge nodes and adjust the initial architecture of the knowledge graph.

[0092] Redefine the mapping relationships between knowledge nodes using graph building tools. Strengthen connection strength and generate new mapping relationships for push content. This adjustment allows recommended content to be more focused on user preferences, improving the targeting of push notifications.

[0093] The system provided by the embodiments of the present invention will be described below. The system described below can be referred to in correspondence with the method described above.

[0094] Please see Figure 5 , Figure 5 This diagram illustrates the structure of a domain knowledge graph construction system based on graph neural networks according to an embodiment of the present invention. The system may include: The data acquisition module 10 is used to acquire basic data in the target domain, perform data analysis on the basic data, and extract knowledge nodes and node paths composed of knowledge nodes from the basic data.

[0095] In this embodiment, the basic data mentioned above is obtained through a pre-established domain knowledge base. The domain knowledge base can be supplemented with the basic data of the target domain stored therein by continuously collecting and summarizing relevant literature, records and interview data.

[0096] The data association module 20 is used to divide a number of target domain node groups according to the contextual association features of knowledge nodes and their matching degree with the target domain. Each target domain node group includes a central node and knowledge nodes that have a logical relationship with the central node.

[0097] By calculating the matching degree between knowledge nodes and the target domain, it is possible to ensure that the selected knowledge nodes are highly relevant to the core needs. This forms a complete logical chain from the association calculation of knowledge nodes to the final division and grouping, ensuring that the knowledge content is highly consistent with the needs of the target domain and laying a solid foundation for subsequent applications.

[0098] The knowledge graph construction module 30 is used to mine the potential connection paths between various knowledge nodes in the target domain node group, construct a knowledge graph based on knowledge nodes, logical relationships, and potential connection paths, determine the core nodes in the knowledge graph, obtain preliminary push content from a preset database based on the core nodes, and determine the push priority of the preliminary push content based on user preferences to obtain priority push content. By analyzing the potential connections between knowledge nodes, we can focus on core relationships and avoid interference from irrelevant information. Core nodes are those with high centrality in the knowledge graph. Centrality allows us to quickly filter out node information that is crucial for knowledge delivery. Acquiring user preferences allows us to capture dynamic changes in user interests in a timely manner, enabling us to adjust the delivery priority based on these preferences, thus achieving prioritized content delivery.

[0099] The domain knowledge graph construction system based on graph neural networks of this invention, after acquiring knowledge nodes, divides them into several target domain node groups based on the contextual association features of the knowledge nodes and their matching degree with the target domain. These target domain node groups are highly correlated with the target domain, ensuring a high degree of fit between the knowledge content and the needs of the target domain. The system then mines the potential association paths between the knowledge nodes in each target domain node group and constructs a knowledge graph based on this. The core nodes in the knowledge graph are determined based on the centrality of the knowledge nodes. Initial push content is obtained from a preset database based on the core nodes. This not only accurately identifies core concepts and key relationships but also enhances the system's adaptability in specific scenarios and improves the application targeting of knowledge nodes. Acquiring user preference content allows for timely capture of dynamic changes in user interests, enabling adjustments to the push priority based on preferences to receive prioritized content. Furthermore, an efficient storage architecture optimizes the response speed for frequently accessed content, achieving intelligent and personalized knowledge push.

[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for constructing a domain knowledge graph based on graph neural networks, characterized in that: The method includes: Acquire basic data in the target domain, perform data analysis on the basic data, and extract knowledge nodes and node paths composed of knowledge nodes from the basic data; the knowledge nodes form node paths through logical relationships. Based on the contextual association features of knowledge nodes and their matching degree with the target domain, several target domain node groups are divided; each target domain node group includes a central node and knowledge nodes that have a logical relationship with the central node; The system mines potential connections between knowledge nodes in a target domain node group, constructs a knowledge graph based on knowledge nodes, logical relationships, and potential connections, identifies core nodes in the knowledge graph, obtains initial push content from a pre-set database based on the core nodes, and determines the push priority of the initial push content based on user preferences to obtain priority push content.

2. The domain knowledge graph construction method based on graph neural networks as described in claim 1, characterized in that: Based on the contextual association features of knowledge nodes and their matching degree with the target domain, several target domain node groups are divided, specifically including: Determine the matching degree between knowledge nodes and the target domain, and retain knowledge nodes whose matching degree exceeds the first preset matching degree; Determine the contextual association features of knowledge nodes, identify the central node based on the contextual association features, and divide the candidate node groups according to the logical relationships of the central nodes; The contextual features of each knowledge node in each candidate node group are fused to obtain node fusion features. The matching degree between the node fusion features and the target domain is determined, and the candidate node group with a matching degree exceeding the second preset matching degree is taken as the target domain node group.

3. The domain knowledge graph construction method based on graph neural networks as described in claim 2, characterized in that: The process involves mining potential connection paths between knowledge nodes in the target domain node group, constructing a knowledge graph based on knowledge nodes, logical relationships, and potential connection paths, identifying core nodes in the knowledge graph, obtaining initial push content from a pre-set database based on the core nodes, and determining the push priority of the initial push content according to user preferences to obtain priority push content, specifically including: Obtain the semantic similarity and context dependency matrix among the knowledge nodes in the target domain node group; Based on semantic similarity and context dependency matrix, and using graph neural network model, the graph neural network model outputs the connection strength between each knowledge node in the target domain node group, and the potential association path between knowledge nodes is mined based on the connection strength. Based on the logical relationships and potential association paths between knowledge nodes in all target domain node groups, an initial framework for the knowledge graph is constructed, and the centrality of each knowledge node in the initial framework is determined according to the connection strength and the number of connections. Knowledge nodes with centrality exceeding the preset centrality are designated as core nodes, and initial push content is obtained from the preset database based on the core nodes; The system acquires user interaction data within a first preset time period, determines the frequency of user interaction with each type of initial push content based on the interaction data, and determines user profile data representing user behavior patterns based on the interaction frequency. The interaction data includes the user's historical data and real-time behavior data. The initial push priority of the content is determined based on the user profile data. The initial push content is pushed to the user according to the push priority, and the user's interaction data in the second time period is obtained. By comparing the interaction data of the first and second time periods, the user's preferred content is obtained, and the push priority is adjusted according to the preferred content. The priority is then determined based on the adjusted priority.

4. The domain knowledge graph construction method based on graph neural networks as described in claim 3, characterized in that: Before the step of designating knowledge nodes with centrality exceeding a preset centrality as core nodes and obtaining initial push content from a preset database based on the core nodes, the method further includes: Adjust potential related paths according to the target domain; the adjustment methods include adding potential related paths and deleting potential related paths.

5. The domain knowledge graph construction method based on graph neural networks as described in claim 1, characterized in that: The method further includes: Priority-driven content is stored using distributed storage, and cache space is allocated to priority-driven content based on its priority; higher-priority content receives more cache space.

6. The domain knowledge graph construction method based on graph neural networks as described in claim 5, characterized in that: The method further includes: Obtain access request logs for each priority level of pushed content, extract access distribution characteristics from the access request logs, and adjust the cache space allocated to priority-pushed content based on the access distribution characteristics.

7. The domain knowledge graph construction method based on graph neural networks as described in claim 6, characterized in that: The method further includes: Based on the adjusted storage space for priority-pushed content, the system retrieves the runtime data of the knowledge nodes corresponding to the priority-pushed content and then distributes access requests for the priority-pushed content according to this runtime data. This runtime data includes response time and resource utilization.

8. The domain knowledge graph construction method based on graph neural networks as described in claim 7, characterized in that: The method further includes: Based on the access request logs of the target domain and the content pushed for each priority, the logical relationships between knowledge nodes are reconstructed, and the initial architecture of the knowledge graph is adjusted.

9. The domain knowledge graph construction method based on graph neural networks as described in claim 1, characterized in that: The process of acquiring basic data in the target domain, performing data analysis on the basic data, and extracting knowledge nodes and node paths composed of these knowledge nodes specifically includes: Acquire basic data in the target domain and convert it into text data; The text data is preprocessed, and then the preprocessed text data is segmented to obtain text fragments; Semantic analysis is performed on text fragments to determine the contextual relevance between them, and text fragments with a relevance exceeding a preset threshold are classified as the same knowledge node; Determine the logical relationships between different knowledge nodes and use these logical relationships as node connection information to connect the knowledge nodes. Based on the node connection information, establish node paths composed of knowledge nodes.

10. A domain knowledge graph construction system based on graph neural networks, characterized in that: The system includes: The data acquisition module allows users to acquire basic data in the target domain, perform data analysis on the basic data, and extract knowledge nodes and node paths composed of knowledge nodes from the basic data; the knowledge nodes are connected by logical relationships to form node paths. The data matching module is used to divide several target domain node groups according to the contextual association features of knowledge nodes and their matching degree with the target domain; each target domain node group includes a central node and knowledge nodes that have a logical relationship with the central node; The knowledge graph construction module is used to mine the potential connection paths between various knowledge nodes in the target domain node group. It constructs a knowledge graph based on knowledge nodes, logical relationships, and potential connection paths, identifies the core nodes in the knowledge graph, obtains preliminary push content from a preset database based on the core nodes, and determines the push priority of the preliminary push content based on user preferences to obtain priority push content.

Images